An ensemble-Kalman filter (EnKF) suitable for mesoscale data assimilation has recently been implemented in the NCAR-PSU mesoscale model MM5. The EnKF is a flexible data assimilation technique designed to use all available information in order to produce the most accurate possible description of the state of the flow. Also it provides the uncertainty in the state of the flow resulting from the uncertainties in various sources of information. Through various observing system assimilation experiments (OSSEs), the MM5-based EnKF is demonstrated to be very effective in assimilating sounding and surface observations with typical temporal and spatial resolutions for an explosive cyclogenesis event with strong error growth at all scales as a result of interactions between convective-, meso- and subsynoptic-scale dynamics. Moreover, the EnKF is found to be quite resilient under various circumstances including ensemble initialization and size, localization, model error, model resolution, and observation error and availability, though imperfect model or imperfect initialization can significantly degrade the filter performance.
Currently, a similar EnKF based on the Weather Research and Forecast (WRF) model is also being developed. We have just begun the OSSEs for the two MCV cases with the MM5-based EnKF and plan to migrate to the WRF-based EnKF assimilation system. We will begin to assimilate in-situ observations of the two BAMEX cases after sufficient understanding of the EnKF behaviors for warm-season mesoscale convective systems has been gained through these OSSE experiments. We believe the proposed EnKF technique will not only maximize the information gained from the direct measure of low-level winds and thermodynamic properties of the boundary layer and low-troposphere but can also induce information of many unobserved state variables (such as those in data sparse area or of vertical velocity) through ensemble estimation of the flow-dependent background error covariance. The integrated data sets and subsequent numerical simulations from the data assimilation will be used to understand the role of MCVs in initiating and modulating convection and its feedback and to assess the predictability of the timing and location of the development and regeneration of convection associated with MCVs.